Choosing a project that matters
The best first project solves a small, real problem for people you can actually talk to. Learn how to find one and cut it down to two weeks of work.
In 60 seconds
Choosing a project that matters
The best first project solves a small, real problem for people you can actually talk to. Learn how to find one and cut it down to two weeks of work.
Start with people, not with AI
- 1
Notice
Write down five problems you see people near you face again and again: at home, school, college, a shop, a society office, a farm. - 2
Ask
Talk to two or three of those people. Ask what they do today, and what goes wrong. Listen more than you talk. - 3
Check the fit
Is the problem about finding information, sorting messages, summarising text or answering the same questions? AI may help. Is it about money, medicine or law decisions? Be very careful, or pick something else. - 4
Cut it down
Keep only the smallest part that would still help. One language, one group of users, one job. - 5
Decide
Pick the idea you can test with real people within two weeks.
Good first projects, and risky ones
- A bot that gives medical advice to patients
- An app that tells people which shares to buy
- A system that decides who gets a loan or a job
- Anything that collects many people's private data
- "An AI for the whole state"
- A FAQ bot for your school library or society office, from notices you are allowed to use
- A helper that flags likely scam messages for your family to check
- A tool that turns long circulars into a short summary with the source
- A study helper that quizzes you from your own notes
- A translator helper for one shop's price list, checked by a person
A simple scoring check
| Question | Why it matters |
|---|---|
| Did real people say they have this problem? | If only you think it is a problem, nobody may use your project. |
| Can you reach the users during the project? | You need them to test it. Your classmates, family or a local shop are easy to reach. |
| Is the data available and are you allowed to use it? | Public notices, your own notes or made-up examples are safe. Other people's private chats are not. |
| Can a first version work in two weeks? | A finished small project beats an unfinished big one. |
| What is the worst thing that could happen if it is wrong? | If a wrong answer could hurt health, money or safety, choose something else or add a human check. |
Working as a group or class
- Give each person one clear role: talking to users, data and content, building, testing, writing the README.
- Agree on one shared place for files and notes, and one short meeting every two or three days.
- Write down decisions. When you cut a feature, note why, so nobody quietly adds it back.
- A teacher can give every group the same project type (for example "a FAQ bot") and let each group choose its own users.
Watch and read more
Lab
A ranked list of your own project ideas, scored on need, reach, data, size and risk.
The problem
- Run the starter. It scores three sample ideas from 1 (poor) to 5 (good) on five questions.
- Replace the sample ideas with at least three of your own. Score each honestly. For "safety", 5 means a wrong answer does little harm.
- Look at the ranking and the warnings. Any idea with a safety score of 2 or less is flagged: change it or drop it.
- Write three sentences: which idea you chose, who the users are, and what you cut to make it fit two weeks.
Practice lab
Edit the code and run it. Python runs in your browser — nothing is sent to our server.
No output. Add print(...) to see values.
You are done when
Hard questions
Try to answer before you reveal. If you can answer these, you understood the lesson.
Q1Your group loves an idea that scores low on "need" because nobody you asked had the problem. What do you do?Reveal
Treat it as a warning, not a detail. Ask a few more people, in the right group, before deciding. If nobody has the problem, the project may be fun but will not get real users, and you cannot test it well. Either find the real version of the problem people do have, or pick another idea.
Q2Why does the scorer give extra weight to need and safety?Reveal
Because those two are the hardest to fix later. You can shrink the size or find other data, but if nobody needs the tool, or if a wrong answer could hurt someone, more building does not help. The weights are a choice: your group can change them, but should write down why.
Questions people ask
What if I cannot find a problem?
Watch for repeated questions. Anywhere people keep asking the same thing (a school office, a society WhatsApp group, a shop counter, your family), there is often a small project. You can also improve one of your earlier lab exercises for a real user.
Does my project have to use AI?
Not every part. Many good projects are mostly simple rules and good content, with AI used for one step, such as understanding the question. If you find that a simple rule or a printed notice solves the problem, that is a useful finding too. Write it down.
Should I work alone or in a group?
Both work. Alone, keep the project even smaller. In a group, split roles clearly and agree on one decision-maker for scope, so the project does not keep growing.
What if my idea already exists?
That is fine. Existing apps may not work in your language, for your town, or for your users. Your version can be smaller and more local. Learning comes from building and testing it yourself.
Lesson test
5 questions. Get 3 right (60%) to pass and complete this lesson.